Real-Time Water-Quality Monitoring for Water Security

ApplicationsGeneral Meeting of the Federal Remediation

Technologies Roundtable, Arlington, VA, Dec. 7, 2005

Ronald J. Baker, Eric F. Vowinkel, and Rachel A. Esralew, U.S. Geological Survey, West Trenton, NJ

Real-Time Water-Quality Monitoring for Water Security

• Can water-quality monitoring increase security?

• What sensors should be used?• How reliable are the available

sensors?• What are the maintenance and

replacement intervals?

“The need to developnear real-time monitoring technologies which would be particularly useful in quickly detecting contaminants in water that has already left the treatment plant for the consumer, has by far the strongest support”

Nearly 100% of experts consider this a high priority.

Existing USGS real-time systems

Objectives of the Overall Water Security Research Program

• To develop a real-time water-quality monitoring system for drinking water safety and security

• To evaluate available sensors for use in such a system

• To install and test the system in water distribution systems

Site Installations

• 5 distribution-system sites established• Free Chlorine, Specific Conductance, pH,

Oxidation-Reduction Potential, and temperature measured at all sites

• Total Organic Carbon and UV/VIS (on loan) at one site

• Additional sites based on model results to be installed (locations optimized for public health protection) – pending logistical issues

Water-Quality Monitoring System

Typical site installation

• Sensor locations: Based on distribution-system modeling– Water utility facilities, pumping stations, homes,

government buildings, hospitals…• Sensors: free chlorine, oxidation-reduction potential, specific

conductance, pH, temperature

Data telemetry and management : Sensors data transmitted to USGS secure webpage using satellite telemetry

Some Realities of Field Work with Sensors

• Often heard questions and comments:– Did anyone bring the manual for this

thing?– Oh, No, not again!– Do we have a spare?– Did I really spend (X) years in college to

do this?– If one more thing goes wrong I will miss

(fill in favorite 9:00 pm TV show).

Example 3: Dissolved Oxygen Sensor Membrane Replacement

Site SW1 Dissolved Oxygen Data

0

4

8

12

08/20/04 08/30/04 09/09/04DATE

DIS

SOLV

ED

OXY

GEN

IN m

g/L

0

5

10

15

20

25

30

TEM

PER

ATU

RE

(C)

Dissolved Oxygen Temperature

Site SW1 Dissolved Oxygen Data

0

4

8

12

16

11/02/03 11/12/03 11/22/03DATE

DIS

SOLV

ED O

XYG

EN IN

M

G/L

Sudden decline in response indicates membrane needs to be replaced.

Gradual decline in response, noticeable as higher-than-normal calibration drift. Membrane needs to be replaced.

Example 4: Turbidity Sensor FoulingWiper keeps particles from building up on window.Removal of wiper causes inaccurately high readings

Site SW 1 Turbidity: Sensor Fouling

-505

1015202530354045

04/13/04 04/18/04 04/23/04 04/28/04 05/03/04DATE

TUR

BID

ITY

AS

NTU

Example 1: pH Sensor that Needs to be Replaced

As probe signal degrades, response in buffer deviates farther from standard value.

Result: Data from 11/04 – 12/04 not accurate, and variability not representative of true system conditions

Site DS3 pH Probe: Calibration Drift

-0.2

-0.1

0.0

0.1

0.2

0.3

0.4

0.5

0.6

9/14/2004 10/24/2004 12/3/2004 1/12/2005

Date

DEV

IATI

ON

FR

OM

pH

7 b

uffe

r

6

6.2

6.4

6.6

6.8

7

7.2

7.4

7.6

7.8

8

pH R

EAD

ING

OF

WA

TER

Calibration Drift at 7 BufferpH Measured (units)

Probe failure recognized

Example 2: Oxidation-Reduction-Potential (OPR) Sensor Performance: Calibration Drift of New

Sensors

0

20

40

60

80

100

120

07/07/04 09/20/04 12/04/04 02/17/05 05/03/05DATE

OR

P C

ALI

BR

ATI

ON

DR

IFT

IN m

V

Site DS 2Site DS 3

New sensors:Response (mV reading) increases after deionizedwater rinse

Platinum electrode builds up an oxide coating which is removed by de-ionized water.

Possible reason:

Over time coating becomes resistant to rinsing, reducing post-cleaning calibration drift

What does this mean? Must account for post-cleaning variability on newer probes!

Summary of Observations of the Sensors Tested

• pH: Mostly stable and accurate. Requires replacement after 1-2 years. Drift noticed over time, electrode requires reconditioning every 4-5 months.

• SC sensor: excellent performance• ORP sensor: very accurate after initial

“break-in” time• Chlorine residual sensor: still beta

testing, needs membrane replacement every 1-2 months. Needs calibration every 2 weeks.

Sensor Performance: Key Issues• The need for sensor maintenance or

replacement is not always obvious by data observation alone:– Data from a nonresponsive sensor may look

reasonable • Sensors must be maintained properly:

– Calibration drift affects data accuracy and precision

– Quality assurance includes assessing and correcting data to reflect drift

– Shorter maintenance intervals and more frequent calibrations reduce chances of significant sensor drift and improve data accuracy

Specific Conductance

0

50

100

150

200

250

300

350

3/10/04 3/13/04 3/16/04 3/19/04 3/22/04 3/25/04 3/28/04

Date

SPEC

IFIC

CO

ND

UC

TAN

CE

IN

uS

Site 1: Distribution SystemSite 2: Source Water

RELATION BETWEEN FREE RESIDUAL CHLORINE AND ORP, DISTRIBUTION SITE 3

-0.2-0.15

-0.1-0.05

00.05

0.10.15

0.20.25

0.3

11/13/2004 11/17/2004 11/21/2004 11/25/2004DATE

FREE

CH

LOR

INE

IN m

g/L

OR

(OR

P/10

0)^2

-0.2

Free Cl (mg/L)

(ORP/1000)^2-0.2

Free chlorine residual influences ORP

Current Work: Characterizing water-quality variability in a distribution system

• USEPA– Conduct controlled laboratory experiments– Select sites based on distribution-system modeling

• Sandia National Laboratories (SNL)– Develop distribution-system and sensor network

models• USGS

– Establish and operate a network of field sites– Collect and manage water quality data – Prepare interpretive reports

• Cooperating Water Utility– Provide and allow access to distribution system sites– Provide distribution system description and model– Support the field effort by preparing water and electrical

connections and drains

New Technologies Being Tested At Site DW3

• First commercial use of chlorine probe in a YSI multiparameter water-quality monitoring system

• Total organic and inorganic carbon analyzer (General Electric)

• UV-VIS spectrophotometer with software for estimating water-quality parameter values and detecting unexpected changes in water quality (S::can Co.)

Trends observed: Most carbon is organic, concentration is within a narrow range (800-1600 ppb), some outliers are present

7/13/2005 7/19/2005 7/25/2005 7/31/2005 8/6/20050

500

1000

1500

Total Organic and Inorganic Carbon in Drinking Water, Site DW3

TOTAL ORGANIC CARBON (ppb)TOTAL INORGANIC CARBON (ppb)

Trends observed: Little variability in 2-day period, twice-daily peak concentrations resembling tidal pattern, “step function” concentration pattern

7/23/2005 7/24/2005 7/25/20051200

1250

1300

1350

TOTA

L O

RG

AN

IC C

AR

BON

(ppb

)Total Organic Carbon in Drinking Water, Site DW3

TOTAL ORGANIC CARBON (ppb)

Specific Conductance, Total Organic Carbon, and Oxidation Reduction Potential - in the absense of Chlorine Residual

1200122012401260128013001320134013601380140014201440146014801500

22-Jul-05

24-Jul-05

26-Jul-05

28-Jul-05

30-Jul-05

01-Aug-05

03-Aug-05

05-Aug-05

07-Aug-05

09-Aug-05

11-Aug-05

13-Aug-05

15-Aug-05

Time

Tota

l Org

anic

Car

bon

(ppm

)

100120140160180200220240260280300320340360380400420440460480500

Spec

ific

Con

duct

ance

(µS/

cm) a

ndO

RP

(V)

Total Organic Carbon (ppm)Specific Conductance (µS/cm)Oxidation-Reduction Potential (mV)

Oxidation-Reduction Potential (ORP) at Five Sites

0

100

200

300

400

500

600

700

07/12/05 07/17/05 07/22/05 07/27/05 08/01/05 08/06/05 08/11/05DATE

OR

P (M

ILLI

VOLT

S)

Columbia LakesStephens LaneChurch RoadBeverlyCamden

Free Chlorine Residual at Five Sites

-0.2

-0.1

0

0.1

0.2

0.3

0.4

0.5

0.6

07/12/05 07/17/05 07/22/05 07/27/05 08/01/05 08/06/05 08/11/05

DATE

FREE

CH

LOR

INE

(mg/

L A

S C

l)

Columbia LakesStephens LaneChurch RoadBeverlyCamden

Specific Conductance at Five Sites

250

270

290

310

330

350

370

390

07/12/05 07/17/05 07/22/05 07/27/05 08/01/05 08/06/05 08/11/05DATE

SPEC

IFIC

CO

ND

UC

TAN

CE

(uS)

Columbia LakesStephens LaneChurch RoadBeverlyCamden

Analysis of Water-Quality Variability

– Spatially• Age of water• Distance between monitoring sites• Type of water (SW, GW, mixed)

– Temporally• 15-minute intervals (or more frequent if

needed)• Hourly• Daily• Weekly• Monthly• Seasonally• Annually

Explanation of the Density Diagram

A “smoothed histogram”, showing the shape of a data set.

X axis: The difference between each measurement (e.g., pH) and the mean of measured values (moving average) within a time increment (15 minutes, 4 hours or 24 hours).

Y axis: Density, or relative frequency of occurrence, of a range of X values

Evaluating density diagrams: Relative magnitudes of density values (not the actual values) are most informative for understanding the shape of the data.

Comparison of Temporal Water-Quality Variability Among 3 Distribution-System Sites

-20 -18 -16 -14 -12 -10 -8 -6 -4 -2 0 2 4 6 8 10 12 14 16 18 20PERCENT DIFFERENCE BETW EEN SC MEASUREMENTS 15 MINUTES APART

0.0

0.2

0.4

0.6

0.8

REL

ATI

VE F

REQ

UEN

CY

OF

OC

CU

RR

ENC

E

SPECIFIC CONDUCTANCE (SC, µs/cm)

Site DW 1 (Northern NJ, surface water)Site DW 2 (Southern NJ, surface water)Site DW 3 (Southern NJ, surface water and groundwater)

OXIDATION-REDUCTION POTENTIAL (ORP, mV): DISTRIBUTION OF DIFFERENCES BETWEEN MEASUREMENTS AND MOVING

AVERAGES OVER 3 TIME INTERVALS

DISTRIBUTION SITE 3

-200 -100 0 100 200DIFFERENCE BETW EEN M EASUREM ENT AND M OVING AVERAGE

0.0

0.1

0.2

0.3

REL

ATI

VE F

REQ

UEN

CY

OF

OC

CU

RR

ENC

E

FIFTEEN M INUTE INTERVALS 95% OF VALUES = ± 5.43 99% OF VALUES = ± 7.14

FOUR HOUR INTERVALS 95% OF VALUES = ± 15.5 99% OF VALUES = ± 21.7

DAILY INTERVALS 95% OF VALUES = ± 29.0 99% OF VALUES = ± 38.1

-200 -100 0 100 200DIFFERENCE BETW EEN M EASUREM ENT AND M OVING AVERAGE

0.0

0.1

0.2

0.3

REL

ATI

VE F

REQ

UEN

CY

OF

OC

CU

RR

ENC

E

FIFTEEN M INUTE INTERVALS 95% O F VALUES = ± 2.72 99% O F VALUES = ± 3.57

FOUR HO UR INTERVALS 95% O F VALUES = ± 11.4 99% O F VALUES = ± 14.9

DAILY INTERVALS 95% O F VALUES = ± 27.8 99% O F VALUES = ± 36.5

-200 -100 0 100 200DIFFERENCE BETW EEN M EASUREM ENT AND M OVING AVERAGE

0.0

0.1

0.2

0.3

REL

ATI

VE F

REQ

UEN

CY

OF

OC

CU

RR

ENC

E

FIFTEEN M INUTE INTERVALS 95% O F VALUES = ± 23.53 99% O F VALUES = ± 30.91

FO UR HO UR INTERVALS 95% O F VALUES = ± 70.02 99% O F VALUES = ± 91.99

DAILY INTERVALS 95% O F VALUES = ± 133.1 99% O F VALUES = ± 174.8

DISTRIBUTION SITE 3

DISTRIBUTION SITE 1 DISTRIBUTION SITE 2

SPECIFIC CONDUCTANCE (uS/cm): DISTRIBUTION OF DIFFERENCES BETWEEN MEASUREMENTS AND MOVING

AVERAGES OVER 3 TIME INTERVALS

DISTRIBUTION SITE 3-50 -40 -30 -20 -10 0 10 20 30 40 50 60

DIFFERENCE BETW EEN M EASUREM ENT AND M OVING AVERAGE

0.0

0.2

0.4

0.6

0.8

REL

ATI

VE F

REQ

UEN

CY

OF

OC

CU

RR

ENC

E FIFTEEN M INUTE INTERVALS 95% OF VALUES = ± 7.55 99% OF VALUES = ± 9.92

FOUR HOUR INTERVALS 95% OF VALUES = ± 22.7 99% OF VALUES = ± 29.9

DAILY INTERVALS 95% OF VALUES = ± 41.3 99% OF VALUES = ± 54.3

-50 -40 -30 -20 -10 0 10 20 30 40 50 60DIFFERENCE BETW EEN M EASUREM ENT AND M OVING AVERAGE

0.0

0.2

0.4

0.6

0.8

REL

ATI

VE F

REQ

UEN

CY

OF

OC

CU

RR

ENC

E FIFTEEN M INUTE INTERVALS 95% OF VALUES = ± 1.43 99% OF VALUES = ± 1.87

FOUR HOUR INTERVALS 95% OF VALUES = ± 5.88 99% OF VALUES = ± 7.73

DAILY INTERVALS 95% OF VALUES = ± 10.8 99% OF VALUES = ± 14.3

-50 -40 -30 -20 -10 0 10 20 30 40 50 60DIFFERENCE BETW EEN M EASUREM ENT AND M OVING AVERAGE

0.0

0.2

0.4

0.6

0.8

REL

ATI

VE F

REQ

UEN

CY

OF

OC

CU

RR

ENC

E

FIFTEEN M INUTE INTERVALS 95% OF VALUES = ± 1.088 99% OF VALUES = ± 1.429

FOUR HOUR INTERVALS 95% OF VALUES = ± 3.674 99% OF VALUES = ± 4.827

DAILY INTERVALS 95% OF VALUES = ± 8.150 99% OF VALUES = ± 10.71

DISTRIBUTION SITE 1 DISTRIBUTION SITE 2

DISTRIBUTION SITE 3

Beverly Free Chlorine Data Through 09/07/05: Cumulative Fractions of Samples Having Consecutive-Concentraton Differences At or Above

T hresholds

0.00001

0.0001

0.001

0.01

0.1

1

0.7 0.4 0.2 0.1 0.07 0.05 0.02 0.01 0Cl2 CONCENTRATION DIFFERENCE THRESHOLD (mg/L)

CU

MU

LATI

VE F

RA

CTI

ON

15 Minute Intervals60 Minute Intervals24 Hour Intervals

Beverly Specific Conductance Data T hrough 09/07/05: Cumulative Fractions of Samples Having Consecutive-Concentraton Differences

At or Above Thresholds

0.00001

0.0001

0.001

0.01

0.1

1

40 25 15 10 6 3 1 0SPECIFIC CONDUCTANCE DIFFERENCE THRESHOLD (Us/cm)

CU

MU

LATI

VE F

RA

CTI

ON

15 minute intervals60 minute intervals24 hour intervals

Specific Conductance Data Through 09/07/05: Cumulative Fractions of Samples Having Consecutive-Concentraton Differences At or Above

T hresholds

0.00001

0.0001

0.001

0.01

0.1

1

0124610152030SPECIFIC CONDUCTANCE DIFFERENCE THRESHOLD (Us/cm)

CU

MU

LATI

VE F

RA

CTI

ON

BeverlyCamdenChurchDelranCol LakesStephen

Beverly Total Organic Carbon (TOC) Data T hrough 09/07/05: Cumulative Fractions of Samples Having Consecutive-Concentraton

Differences At or Above T hresholds

0.001

0.01

0.1

1

200 100 80 60 40 30 20 10 0TOC DIFFERENCE THRESHOLD (ug/L)

CU

MU

LATI

VE F

RA

CTI

ON

16 minute intervals60 minute intervals24 hour intervals

Long Term Program Objectives

• Distribution system sites will serve as practical test beds for sensors and EWS components in the field

• Potential model calibration effort using a tracer in a sub-area of the distribution system

• Support collaborative efforts of the EPA/NHSRC and Sandia National Laboratories (SNL) to characterize water-quality variability for distribution systems

What will be learned?

• Use of hydrologic and water-quality model to select sensor location

• Natural variability of water-quality characteristics in a distribution system

• Capabilities, limitations and maintenance requirements of sensors

Summary• Overall Program Goal: Design and test a

network of water-quality sensors for drinking water safety and security

• Sensors have been selected, site design and telemetry were developed, and water-quality data are being collected at 6 sites

• Real-time data will be used to determine the temporal and spatial variability of water quality in a distribution system

• Most important question remaining: Will introduction of contaminants of concern cause detectable changes in water quality?